Studies of the salivary cortisol awakening response (CAR) may be confounded by delays between waking in the morning and obtaining the 'waking' salivary sample. We used wrist actigraphy to provide objective information about waking time, and studied the influence of delays in taking the waking sample on the CAR. Eighty-three men and women (mean age 61.30 years) who were referred to hospital with suspected coronary artery disease were studied. Saliva samples were obtained on waking and 15 and 30 min later. The mean interval between waking defined by actigraphy and reported waking time was 6.12+/-(S.D.) 14.8 min, with 55.4% having no delay. The waking saliva sample was obtained an average 5.78+/-15.0 min after self-reported waking, and 12.24+/-20.3 min after objective waking. The waking cortisol value was significantly higher in participants who had a delay between waking and sampling >15 min (mean 14.46+/-6.34 nmol/l) than in those with zero (mean 10.45+/-6.41 nmol/l) or 1-15 min delays (mean 11.51+/-5.99 nmol/l, p=0.043). Cortisol did not increase between 15 and 30 min after waking in those who delayed >15 min. There were no differences in CAR between participants with zero and 1-15 min delays from objectively defined waking to reported sample times. A small proportion (14.7%) of participants who did not delay saliva sampling showed no increase in cortisol over the 30 min after waking. These CAR nonresponders did not differ from the remainder on sleep patterns, waking time, clinical or medication characteristics, but were more likely to be of higher socioeconomic status (p=0.009). We conclude that long delays between waking and obtaining 'waking' cortisol samples will lead to misleading CAR results, but that delays up to 15 min may not be problematic. A small minority of individuals do not show a positive CAR despite not delaying saliva sampling after waking.